This is not about "House Rules x RAW", however.
I would guess most groups would quickly accept that the DM can include new monsters to the setting as desired, but this is an assumption that often gets ignored when discussing old school D&D.
For example, I always found that "immunity to ghoul paralysis" was a silly trait to give a class or creature because it is so specific. But if you are only using Moldvay's Basic - and not even expert - it could be an interesting advantage, since the total number of monsters is low. Still, people keep using it even when playing some form of UD&D.
Same for the cleric "turn undead" tables that include the NAMES of the creatures turned instead of their HD. If the number of undead are limited, this makes perfect sense; otherwise, it looks strange and impractical.
(A more extreme example is the blink dog and its hatred of displacer beasts. B/X tell us almost nothing about these creatures, but they attack each other on sight. How often will that happen? If you're playing UD&D, almost never).
There are other aspects of old school D&D that seem to be remnants of this limited mindset. For example, clerics and mages cannot use sword, which is important if most magic weapons are swords, but becomes less important if you have several magic maces or daggers.
Another example I've been struggling with lately is spells. I've been running a game slightly based on B/X, which does not contain "counter spell", exactly. But now I'm introducing an NPC for another system that has this spell. Is it fair to my caster that he didn't get to choose it?
(My solution for this is: if my player shows interest in Dispel Magic, I'll let it function as a counterspell. Fortunately, he hasn't got it so far).
Personally, I was always attracted to this "Unlimited" take on RPGs in general. I'm willing to add new monsters, spells and even the ocasional laser guns to my games, and always tried to accommodate every character concept the players suggest (although now I'm tempted to go mall-human for the next campaign).
But there are advantages of the "limited" perspective - it allows players to get more familiar with existing monsters, spells, etc., for example.
In fact, having fewer monster types reinforces their significance rather than diluting it.
If every single monster, spell or item the PCs find is completely new, they can never learn anything except trough direct contact. There is no room for extrapolation, generalization, etc.
I have a similar feeling on classes. Yes, I like paladins, assassins, warlords, druids, avengers, and monks. But rather than having a dozen classes, I prefer having FOUR: fighter, mage, thief, cleric, each with a few variations. And I think the AD&D bard - a class that works in a completely different way from other classes - is an unnecessary mess.
(Maybe I'll reduce it to THREE classes for my Sword and Sorcery game, ditching the cleric).
Even 5e seems to have problems with this (correct me if I'm wrong; I don't play 5e anymore). For example, the "Staff of Charming" requires attunement by "a bard, cleric, druid, sorcerer, warlock, or wizard". But what if I'm using a 3rd-party class, or even the very popular artificer? Are they automatically excluded? Or do including a class requires rewriting all magic items like that?
As you can see, seeing classes through a limited scope simplifies some aspects of the game.
(This is partly what Old School Feats is about, BTW - no extra classes but many additional options. You can look at the free previews to see how I treat fighters, for example: a framework to create warriors, paladins, rangers and warlords).
In short, my favorite approached to D&D is having unlimited choices within a limited framework that works as a common language between players and DM.





